@chilon249 i got it working but that was definitely not it lol

I think the smaller file is a "pruned" file, which means all unnecessary data has been removed. For making images it's functionally identical to the larger file but if you want to combine two checkpoints together you would want the larger one.

@tombot Okay, so I don't really need both, then. Cheers.

This one actually does know stuff like worst quality, but that's because it bakes in this Lora which uses that as a keyword.
https://civitai.com/models/127533/fluffyrock-quality-tags

skill issue

@exfaqnksxazgknad6232 Skill issue? yeah ok mate. What ever you say. So over compensating by prompt stuffing with words the models aren't trained on helps does it? There is an actual difference in quality when you stuff that "Best Quality" or "best lighting" in there? That sounds like an ignorance issue. Try learning what your model knows and putting prompts in there it will recognise.

if you want an actually good model, one that is capable of different styles and art types, you gotta expect complexity and more prompts. this is easily the most flexible and diverse model of its type on the site, it just needs you to be specific. no 5 word prompts for this one. sure, theres no hand holding like a lot of the other models out there, but damn can you create some good looking results.

cope

@dillon101 "cope" is what people say when they want to sound like they know better but in reality just don't like what someone has said. You sound more ignorant than I sounded with my rant. I may have been frustrated and overwhelmed but you just sound like an ignorant troll who has nothing of quality to add.
You just go and "cope" with yourself in your little corner my dude.

@griddo I appreciate you offering some constructive help, instead of just snide ignorant comments. I will take another look at it all with a fresh set of eyes. Thank you. 

@CaptainKokomo To me, Cope = Deal with it. Your comment sounded like you really didn't care about setting it up, you just wanted results. Instead of asking for help, you end the comment saying you are giving up and going to another option. That's probably why you have so many thumbs down.

As far as I know, all you need to get this model working is to download it, put the model in the webui, then type a prompt. It should work just fine out of the box.

Sorry for being a jerk.

It certainly can!

Found how to do it with controlnet; in the process it shows me a nicely colored pic, but as a finished result it gives me inverted negative colors: https://imgur.com/a/qcUfLDx
does anyone know how to deal with this?

@jeluch in the preprocessor drop down for your controlnet node, try changing it to "invert"

I'm just as stumped as you are on this one... I managed to make it happen as I was hoping for exactly once with "genital outline", but even between seeds it broke again.

@sansmia accidentally found this https://civitai.com/models/50030?modelVersionId=57282
works much better than basic prompts, but not perfect.

It's a difficult tag to get to work and depends on just how much you want to see. 'detailed bulge' is one you could try but its representation in most models is small.

'penis outline' is another one but you will have to wrestle with this especially if you upscale as the outline will quickly just become... well I think you get it.

finally 'translucent clothing' or/and 'translucent underwear' can help, this may need to be coupled with 'wet clothes' also.

there's no real magic bullet for it however. ^^

I have some examples under my profile (eastern dragon by indigowing is probably the one I like the most) but they were done with a custom model and took a lot of trial and error to get.

https://civitai.com/models/57731/eastern-dragon?modelVersionId=62163 - link to indigowings eastern dragon lora.

@chilon249 Do I put the folder somewhere in stable diffusion like in the models? Sorry if this is a stupid question

@chilon249 thank you

@chilon249 also what about the other folders that aren't wildcards?

I'm even running into this issue with the loras that are meant for this model, can anyone give me some pointers?

This is a problem with the stable diffusion itself. Since he was mainly trained on 512X512 images, when the resolution increases above 1000, he often begins to duplicate images. This occurs in all stable difusion models.

It's better to generate a low resolution and then upscale it

can you make a wildcard version of that? or know a wildcard of artist? im triying to use a randomizer of artistic styles

Incredible. I was creating my own in house version of basically this, and now I can stop. You saved me a lot of time. Thanks for sharing

@soulwater2017765 I haven't created wildcards for all artist tags, but I have created wildcards in the anime style.

@sansmia thank you! It was worth spending a lot of time on.

@SuinSuin oh dont worry i understand but thanks this will be helpfull to do the wildcard i will try to make it by my selft if you let me use it this :3 and thaks you

It would be great also if you reupload some old LORAs (if you have them), because the new ones are pure garbage.

@sasisa Sorry, i don't have those

If you are talking about generation on the site, then there was a huge red sign saying that the generator is temporarily not working

There were Loras for that type of thing on this website (I distinctly remember adding them to my ignore list), but due to the terms of service on this website they were all deleted. You might have luck finding them elsewhere though. Aside from that you could just go to e612 and see if there are any specific tags to add into the positive or negative prompts to get what you want.

This isn't my thing but I'm not above melting my eyes to help someone figure out a prompt.

The biggest problem you're going to face is a lack of training data to guide the model to the solution you seek. 99% of bulges is a front bulge so even if you get one, it tends to be in the wrong spot. I second tombot and suggest looking for some loras that could help you out. That said, you can throw enough word soup at it to give it to give you a decent approximation. Try adding the following to your prompt and see what happens:

(wet feces:1.1), (soiling, soiled underwear, dirty underwear:1.3), (feces in underwear:1.2), (leaking feces:1.2), shitting pants, pooping pants, butt bulge, (brown underwear, dirty underwear:1.25).

That's the best my limited look into the topic yielded. Now if you'll excuse me, I must drown myself in cute things for a while

@sansmia What a champ! Someone give @sansmia an award for really going above and beyond! 🏅

@Nono2461 : It seems like you don't let others' values bother you. Good for you! You realize it's a rather niche kink, and that others will judge you for it, yet you still accept it. That's great. Despite what most people would believe about you, I think that's a sign of good psychosexual health.

The only thing that makes one kink more "acceptable" or "pleasant" than another is that there are more people who have it. Popularity becomes normality. But normal doesn't mean better. Scat as a kink is quite unpopular, so it's usually seen as abnormal and "gross"... perhaps even unhealthy. How many people enjoy oral sex? It's so common that we don't even think of it as kinky, but popularity is the only difference between that and other kinks.

For those who have 👎 you: Everyone has kinks, and people don't really choose their kinks. Their kinks choose them. I made a choice about 15 years ago to accept one of my kinks instead of running from it. But I never made a choice about whether to be turned on by it. It's my hope that one day, talking about kinks will no longer be taboo, that everyone feels comfortable about what turns them on, and that no kink is considered more or less "okay" than anothers.